by [author_box layout=”inline” show_title=”false”]
You can find an audio recording of this post at the bottom of the page.
Let’s talk about ”background knowledge.” The term refers to things a child knows about the world–about people, events, and situations; categories such as animals and clothing; topics such as animal habitats and how plants grow. These are things we use language to talk and write about. This knowledge provides the “background” or context for understanding a text.1
Unsurprisingly, there are differing views about the role of background knowledge in early reading.
One side is represented by theories such as the Simple View of Reading, which states that early reading has two components: knowledge of the written code (the part that is unique to reading), and knowledge of language (already known from speech).2 The Simple View is an observation about conditions at the onset of reading. By learning about print readers gain a portal to all they have already learned from using spoken language. Background knowledge is implied in this theory (because language is used to talk about things in the world) but not treated as a principal component of beginning reading. Theories that emphasize the primacy of learning about the printed code in early reading are sometimes said to emphasize “skills” rather than “meaning” or “literacy”.
An opposing view assigns far greater weight to background knowledge in learning to read. Proponents of this view emphasize “literacy” and “meaning” over “skills.” Reading is seen as “an active, constructive, meaning-making process.” The words on the page are only the starting point for generating the interpretation of a text, which varies as a function of a reader’s personal experience, knowledge of the world, and culturally-determined beliefs about the purpose of reading. Skilled reading is seen as a heavily top-down, knowledge-driven process: given sufficient knowledge of language, topic, and genre, a child can accurately anticipate words in texts and only has to sample the letters lightly to confirm their guesses. The instructional implication is that children need to learn how to combine different types of background knowledge (or “cues”) to recognize words, doing away with the need to dwell on “skills” such as gaining familiarity with the spellings of words and their pronunciations.
We think that these seemingly opposing views can be reconciled by taking a cognitive perspective, which addresses the mental operations involved in reading, and a developmental perspective, which addresses how behavior changes as children acquire knowledge over time. Let’s look at how.
Take Me Out To “The Baseball Study”
When people talk about background knowledge they invariably bring up Recht and Leslie’s 1988 Baseball Study.3 Many people have heard of the study, but how many have actually read it? Relying on the second-hand, simplified version of a research study can lead to instructional practices that are meant to be consistent with the “science of reading” but are actually only loosely related to it. The Baseball Study illustrates that concern, so let’s take a closer look at it.
The article reports a small behavioral study in which middle school students read a passage about a half inning of a baseball game. Participants then performed several tasks to assess their comprehension of the passage and memory for specific statements. The researchers also obtained separate measures of the students’ reading ability and their knowledge of baseball (the “background knowledge” relevant to the passage). Participants were sorted into four groups based on these measures:
High reading skill
High baseball knowledge
Low reading skill
High baseball knowledge
High reading skill
Low baseball knowledge
Low reading skill
Low baseball knowledge
We want to know how the individuals in the 4 groups performed on the comprehension and memory measures. The surprising finding that accounts for the study’s notoriety is that performance was determined by background knowledge not reading skill. The participants with better knowledge of baseball did better than those who knew less about it. But, that was true for both good and poor reader groups. Conclusion? The study seems to indicate that in comprehending a text, background knowledge plays a more important role than reading skill. Knowing a lot about a topic can compensate for differences in ability to read. This interpretation is consistent with the top-down, guessing approach we mentioned above. It reinforces the idea that readers don’t have to pay close attention to the text if they have sufficient background knowledge.
If you want to get deeper into the “science of reading,” you want to learn to think like a reading researcher. That means looking at how a study was conducted and whether the data actually support the authors’ conclusions. Sometimes they don’t.
We’ve created a bar chart that makes it easier to understand the results of this experiment. We want to assess the impact of two factors, Reading Ability (high vs. low) and Prior Knowledge (high vs. low), on performance. We’ve plotted the data from a measure of participants’ memory (“verbal recall”) for content of the text; the results are very similar for the other performance measures. The figure shows each group’s mean performance on the task (it’s the average of the scores of all the individuals in that group). The lines on each bar (called “error bars”) are the standard deviations of the means, which indicate how variable the scores are in the group. As a rule of thumb, the means between two conditions only differ reliably if the error bars do not overlap.
This is a 2 x 2 design: two factors (reading skill, prior knowledge) each of which has two levels (high, low). The data analyses are not complicated. Was there an effect of reading skill (high better than low), ignoring baseball knowledge? Was there an effect of baseball knowledge (high better than low), ignoring reading ability? Then we ask about the combination of the factors. That means asking whether any of the four conditions differ from each other. Did the two reader groups with high baseball knowledge perform alike or differently? For the two groups with low baseball knowledge, did higher reading ability allow the participants to perform better? And so on.
The results show a big effect of Prior Knowledge: people with better knowledge of baseball performed better on this measure. In both reader groups, the blue bar differs from the red. The other main finding is that the results are very similar for the high and low reading ability groups. There is no effect of reader group in this study: the high and low reading groups did not differ in performance. Eyeballing the data, it looks like the means in the two low reading ability groups are lower than in the two high reading groups, but these differences are not statistically reliable. The small numerical difference between the means in the two high prior knowledge groups (blue bars) is not statistically reliable: the standard deviations overlap. Each mean falls within the standard deviation of the other group. Informally, the two means do not differ given the “margin of error”. The same is true for the two low knowledge conditions: it looks like the low reading ability/low prior knowledge group performed worse than the high reading ability/low prior knowledge group, but the means do not differ statistically.
Finally, the interaction between the two factors is also not significant. The effect of baseball knowledge was the same for high and low reading groups.
What have we learned from the study? Participants who knew more about baseball performed better than those who knew less. This finding is obvious. It doesn’t pass the “Grandmother test,” which is, could your grandmother guess the result correctly? You ask your grandmother, “if people read a story about a baseball game, who will remember more about it later: people who know a lot about baseball or people who don’t?” Your grandmother says, “the people who know more about baseball. This is what you study?”
What about the fact that the impact of baseball knowledge was the same for better and poorer readers? Does this mean that text comprehension is mainly determined by background knowledge–knowledge of the topic? That background knowledge can compensate for differences in reading skills such as recognizing and understanding words? Does the study support the top-down guessing theory?
Folks, the study says exactly nothing about the relative importance of background knowledge and reading skills. The authors wanted to compare good and poor readers but didn’t succeed in doing so. There are no effects of reader group in this study. The procedures used to select participants did not yield groups that differed in reading skill. The study therefore does not show anything about the effect of reading skill or the interaction of reading skill and prior knowledge. The experiment would have yielded the same results if the texts had been read aloud to the participants–if there were no reading involved at all.
The authors made the Stats101 error of interpreting null effects–the absence of a reliable statistical effect of reading skill or interaction between reading skill and prior knowledge–as providing positive evidence that reading level didn’t matter, only background knowledge. That is not OK. The results are equally consistent with the conclusion that the groups didn’t differ in reading skill, or at least enough to have any measurable impact on the performance in this experiment.
The study didn’t yield a valid test of the impact of reading skill because of choices that were made in designing and conducting it. First, the differences between the good and poor readers were small because the authors excluded certain poor readers: “We selected low-ability subjects who scored above the 30th percentile in vocabulary to avoid word-recognition problems.” The weakest readers–the ones who really struggle with reading words–were excluded. Moreover, the simple tests that were used to assess reading skills have their own imprecision (margin of error). The net result is that the good and poor readers did not differ substantially in reading ability. The problem could have been avoided by choosing participants from a broader range of reading skills.
Second, there were only 16 participants in each condition. How many participants should there be in a study? Is 16 a large number or small? That depends on what’s being measured. Given the subtle, at best, differences between the good and poor readers, this number of participants does not yield enough data to detect differences between groups, should they exist. Statisticians would say the study lacked sufficient power to detect effects of reading skill. The study did have sufficient power to detect the effect of background knowledge because it is huge: readers who don’t know much about baseball will have great difficulty remembering what they read. Big effect: doesn’t take many observations to detect. Small effect: you’re gonna need a bigger sample.
Bottom line: though widely cited, the study is fatally flawed and the conclusions do not follow from the results. The study says nothing about the relative importance of background/prior knowledge versus reading skill. Don’t make the mistake of assuming that because a study is published the conclusions are correct.
There’s a better way to look at the roles of background knowledge and basic reading skills, and it doesn’t require comparisons between good and poor readers. In a classic study by Bransford and Johnson, high school and college students heard a passage and then were asked how easy the passage was to comprehend and to recall as many ideas as possible.4 Sounds similar to the Baseball Study, but try it yourself:
If the balloons popped, the sound wouldn’t be able to carry since everything would be too far away from the correct floor. A closed window would also prevent the sound from carrying, since most buildings tend to be well insulated. Since the whole operation depends on a steady flow of electricity, a break in the middle of the wire would also cause problems. Of course, the fellow could shout, but the human voice is not loud enough to carry that far. An additional problem is that a string could break on the instrument. Then there could be no accompaniment to the message. It is clear that the best situation would involve less distance. Then there would be fewer potential problems. With face to face contact, the least number of things could go wrong.
How did that go for you? What was happening in the passage? You likely know all the words and yet didn’t understand what was happening. It was the same for participants in the experiments. However, the results were different when the study was repeated with one change: participants saw the following picture before hearing the passage (hover your mouse over the box below to see the picture).
Suddenly, the passage makes more sense. The participants who saw the picture first showed better recall and comprehension than the participants who didn’t see the picture or even those who were shown the picture after hearing the passage.
Here’s another passage to try:
The procedure is actually quite simple. First you arrange things into different groups depending on their makeup. Of course, one pile may be sufficient depending on how much there is to do. If you have to go somewhere else due to lack of facilities that is the next step, otherwise you are pretty well set. It is important not to overdo any particular endeavor. That is, it is better to do too few things at once than too many. In the short run this may not seem important, but complications from doing too many can easily arise. A mistake can be expensive as well. The manipulation of the appropriate mechanisms should be self-explanatory, and we need not dwell on it here. At first the whole procedure will seem complicated. Soon, however, it will become just another facet of life. It is difficult to foresee any end to the necessity for this task in the immediate future, but then one never can tell.
Got it? What if I told you the passage is about washing clothes? Read it again—suddenly it makes sense! This experiment, along with the others in the paper, suggest that providing enough context about the topic of a passage is critical to comprehension and recall.
However, it does not mean that knowing the topic is enough for comprehension. Just as with the Baseball Study, if the students in the Bransford paper didn’t know the meanings of the words in the passages, telling them the topic wouldn’t help them very much. Without the ability to both read and understand the words presented to you, background knowledge or context is pretty useless.
Implications for Instruction?
These studies and the research that has followed highlight the importance of relevant background knowledge and linguistic context for reading comprehension.5 Children’s reading difficulties can indeed result from lacking such knowledge, rather than problems with recognizing words. This issue arises in many real-world situations. A math word problem may describe a situation that is unfamiliar to some readers because it is outside their culture or experience–like a question that asks about health club membership fees. The same holds for standardized assessments of reading comprehension: children read texts that describe situations, events, objects, and people who are more or less familiar given their backgrounds. The materials are written as if children share certain types of knowledge, but that assumption is often untrue.
Background knowledge is important, and overlooked in formulations such as the Simple View of Reading. It doesn’t follow, however, that having background knowledge is sufficient, or does away with the need to attend closely to the words in a text. The Bransford studies show that a good reader who lacks the relevant background knowledge will comprehend poorly. But numerous studies also show that texts are hard to comprehend if people can’t read the words (ask your Grandma). The goal of instruction is to get children to the point where they recognize words quickly, accurately, automatically. Reading then allows the child to expand their knowledge of language and the world, gaining additional practice reading words and comprehending texts.
This is a developmental perspective. It is not just that both basic skills (word recognition) and background knowledge are important for reading comprehension–though they are. It is that they develop, in different ways, at different time scales. A recipe contains a list of ingredients, but making the dish is a series of steps, some overlapping and some sequential. Moreover, the components are not independent: background knowledge affects reading which depends on word recognition which affects which background knowledge is relevant which depends, in part, on reading varied texts for varied purposes. Basic skills and background knowledge cannot be disentangled from each other–they are intrinsically intertwined and therefore must both be the focus of instruction.
— Mark Seidenberg & Molly Farry-Thorn
<a href="https://seidenbergreading.net/wp-content/uploads/2023/03/85323-some-context-on-context.mp3" target="_blank" role="button" rel="noopener">
(this will open the audio in a new page where you can click the three dots and then click download)
5. How Much Knowledge Is Too Little? When a Lack of Knowledge Becomes a Barrier to Comprehension, O’Reilly, Wang, & Sabatini (2019)